An Intelligent System for Detecting Image Spam based on Exploration of Text Information Embedded into Images

Abstract

Spam mails act as a immense hazard to communication security as it leads to phishing or virus attacks that could harm the user accounts and the organizations by exposing confidential information.With these inferences, the present research attempted to accurately detect image spam e-mails which have always been a topic of great research in data security. Detection of image spam mails is divided into two separate components- character segmentation and character recognition. While the former segments individual characters, the latter overcomes the issue of blocked texts and not well surveyed. The final phase of the work is a complete image spam detection system with the two proposed works built to detect spam messages. Various techniques have been proposed to tackle this setback and the principle of this paper is to evaluate and review several algorithms, talk about benchmark statistics or data, show appraisal of performance followed by potential direction for future work.